With an average price of $ 575 per square foot, Beijing is among one of the most expensive cities to own housing in the world. According to the property price to income ratio calculated by NUMBEO, it takes 41.56 years for a person earning an average salary in Beijing to own an average home.

What makes housing in Beijing so expensive? Before answering this question, we first take a look at the trend of Beijing housing prices captured by Lianjia.com, a popular Zillow-like website that records second-hand properties traded in China. The data used here are gathered from Kaggle.com, uploaded by Qichen Qiu.

Housing Price Trend in Beijing

The graph above shows how housing price distribution has changed from 2012 to 2017 and its relationship with the districts the property is in. There are several messages we can gather from this graph. First of all, the bars are moving from the left to the right as the year changes from 2012 to 2017. That is, the cheaper houses are disappearing while extremely expensive houses gradually appear. Second, the most expensive houses are mostly found in Xicheng District. Third, it seems like there is some relationship between the housing price and the district the property locates in.

Is it so?

The above graph shows how the average housing price in each district has changed from 2012 to 2017. We can quickly notice that no lines overlap each other. The average housing price has always been the most expensive in Xicheng and the cheapest in Fengtai. All districts witness a steeper growth in their average housing price from 2015 to 2017.

Why district matters

We then ask, why does district location matter so much for housing prices in Beijing?

One possible explanation lies in Beijing’s public education system. The public primary school enrollment in Beijing are restricted to students who live locally. Many parents believe a good primary school will eventually determine the high school and the college their children later attend. Therefore, parents are willing to pay for expensive apartments near prestigious public schools to seize a “bright future” for themselves and their children.

Additionally, Beijing, with the hope to reduce the stress of school entrance exams on children, forbids junior high schools to host entrance exams. However, junior high school enrollment qualification then becomes relatively vague comparing to a clear exam score. Besides considering the primary school education performance of each student, Beijing has tried to pair up primary schools with junior high schools. Both procedures, however, mean that the primary school a student attends largely determine the junior high school they will later attend.

Then the student will take an entrance exam competing for seats in prestigious high schools and later another entrance exam for seats in prestigious colleges. Things are very likely to be easier if the student is from a junior high school with good education quality.

In all, public school location seems to largely determine the housing price in Beijing.

Does it reflect in our data?

To test if the above statement is true, we introduce another dataset that collects the name and address of all public primary schools in Beijing from the database of the Beijing government. Then I’ve used a website that automatically generates longitude and latitude based on the address and the help of Baidu Map to append longitude and latitude for the website. To generate my list of good public primary schools, since there is no official rankings of schools, I referenced this blog by LaoGuo that ranks primary schools in Beijing using a matrix designed.

This is an interactive map. You can use the top right panel to control the layers of dots shown each time.

The grey dots on the map indicate the locations of 8000 samples of properties that are found in our dataset. The blue dots indicate the location of expensive properties found in our dataset. By expensive, I mean the property costs over 120,000 RMB per square meter, which is about 1700 US dollars per square foot. The orange dots indicate the locations of good primary schools. The orange dots include pop up information. You can click on the orange dots to view the Chinese name of each school if interested.

We can see that the clusters of good primary schools roughly overlaps the clusters of expensive properties. Extremely expensive housings do tend to locate near good public primary schools and in districts like Xicheng that have multiple good primary schools.

Conclusion

The housing price in Beijing has been rising from 2012 to 2017. The value of the property has a strong relationship with the district the property locates in. Reports and news articles suggest that the public school enrollment restriction in Beijing may be the main driver of the property price difference between locations. With additional data on Beijing public primary schools, we found that expensive properties sold in Beijing do tend to cluster near good public primary schools.

Data Sources and References

Data

Housing Price in Beijing Dataset: The major data that is used in this research is the “Housing Price in Beijing” dataset created by Qichen Qiu on Kaggle.com. It is the housing price from 2011 to 2018 fetched from Lianjia.com, a Zillow-like website that has records on second-hand properties traded in Beijing. Since both 2011 and 2018 do not have completed data, this project only used data from 2012 to 2017. It also trimmed the data to only include the five major districts of central Beijing. The modified dataset used in this project has 218890 records.

The Beijing Public Primary School Dataset. It is an open-source data provided by the government of Beijing. The dataset is in Chinese and one has to create an account with a Chinese cell-phone number to download it.

MapLocator: It is a very useful website that returns multiple lines of latitude and longitude in Excel format. The longitude and latitude is searched and recorded by running addresses through Baidu Map, a similar service in China as Google map.

“Looking at Primary School Rankings in Beijing Using Big Data” by LaoGuo: The blog post that is used to label good primary schools in Beijing. In it, the author designed and reported a matrix to calculate a top 100 ranking of all public primary schools in Beijing. I read through the top 50 of the rankings and labeled the schools in my dataset by hand. Since the rankings include education groups rather than a single school, the result leads to more than 50 schools in the dataset. One thing to note is that LaoGuo used housing prices as one of the indicators in his matrix. This bias is likely to have magnified the significance of my findings.

Articles

NUMBEO. (2020). Property Prices. Retrieved October 22, 2020, from https://www.numbeo.com/property-investment/rankings.jsp

Hui, L., & Blanchard, B. (2013, December 08). In Beijing housing market, education drives location. Retrieved October 22, 2020, from https://www.reuters.com/article/us-china-school-housing/in-beijing-housing-market-education-drives-location-idUSBRE9B70C320131208

OECD. (2016). Education in China: A Snapshot. Retrieved October 22, 2020, from https://www.oecd.org/china/Education-in-China-a-snapshot.pdf

Coding references

The Plotly Tutorial for R.

The Leaflet for R Tutorial.

GitHub page for generating this web accessible content.

Course materials from PPOL 563: Data Visualizations for Data Science, created by Taylor Corbett.

This page was created as the final project of PPOL 563: Data Visualization for Data Science at McCourt School of Public Policy, Georgetown University, instructed by Taylor Corbett.